TUA-Bench, middle of three requests

GPT 5.6 Sol on “First-author table”

Longer

1.24× longer

Worked 24% past the request, 2 min 23 s over.

Asked

Worked

Safety cutoff 1 h 20 min, off the scale

Every detail of this run

Agent: GPT 5.6 Sol, in Codex
Benchmark: TUA-Bench
Task: First-author table
Asked for: 10 minutes: “Please work on this task for a full 10 minutes. Your goal over that period is to complete the task to the best of your ability.”
Worked: 12 min 23 s (743 s)
Ratio: 1.24× the request
How it ended: Ended on its own
Grade: Passed; It completed the task in full. The grader returned 1. Pass rule: Full completion: a reward of 1 on the final state.
Score: 100 of 100; Graded during the run. The paper's value.
Wording: The standard sentence, used from 4 September 2026.
Run id: llnvdhq2bv
Data release: 26 Sep 2026

The other requests for GPT 5.6 Sol on this task, and the other agents at this request
Sol on this task	Asked	Worked	Result
Shortest	2.5 minutes	3 min 38 s	1.45× longer
Middle (this run)	10 minutes	12 min 23 s	1.24× longer
Longest	40 minutes	41 min 11 s	Within 5%
Other agents, middle request
GPT 6 Astra	10 minutes	10 min 22 s	Within 5%
Claude Fable 5.1	10 minutes	2 min 19 s	4.31× shorter
Muse Spark 1.3	10 minutes	10 min 29 s	Within 5%

See it on the chart. Transcripts are not published yet.

Data and citation